Runtime Failure Detection

摘要

We propose a model-driven methodology to generate and automatically deploy runtime failure detectors that enable self-healing software to deal with functional failures. The methodology rests on addressing key research challenges: (1) Identifying classes of properties specified in end-user requirements, which can be associated with typical integration failures. (2) Designing runtime efficient detectors for these failures. (3) Defining mappings of entities between the domain of end-user requirements and the design model.